Scholarlink

<p>The increasing demand for high functionality and dimensional precision in manufacturing has intensified the need for advanced production technologies for corrosion-resistant stainless-steel fasteners. Conventional nut production relies largely on cold forging or machining; however, stainless steels such as SUS304 and SUS316 exhibit low formability at room temperature, resulting in machining-dependent processes with high cost and low material yield. To address these limitations, this study develops an intelligent warm-forging process system capable of producing stainless-steel nuts with improved precision and process stability, based on real-time multi-sensor data and AI modeling. Key process variables—including Material Heating Temperature, Wedge Stroke, Material Feed Length, Die Temperature, Product Temperature, Forming Force, and dimensional responses such as Head Outer Diameter, Body Outer Diameter, Product Height—were continuously collected from an operating production line. Statistical analysis, correlation analysis, causal inference, and machine-learning models were applied to quantify the effects of these variables on dimensional variation and defect rates. The resulting models were integrated into a smart-manufacturing execution system (MES) that supports real-time monitoring, AI-based defect prediction, and parameter guidance. A formal causal analysis using directed acyclic graphs (DAGs) and the DoWhy framework was conducted to distinguish causal from correlational relationships among process variables. Model generalization was assessed through temporally ordered train–test splitting and 5-fold cross-validation, and a systematic error-source analysis was performed to identify unmeasured confounders limiting dimensional prediction performance. The proposed framework demonstrates the feasibility of AI-assisted warm forging, enabling more consistent product quality, reduced deviation, and enhanced process reliability, thereby contributing to the digital transformation of Korea’s core manufacturing sectors.</p>

错误：搜索内容不能为空，请输入英文关键词
错误：关键词超出字数限制，请精简
高级检索

AI-assisted optimization of warm forging using real-time MES sensor data

  • 期刊论文

  • 出版模式：

    开放获取

  • 发表日期：

    2026年7月31日

查看全文
  • Soonkwan Hwang,
  • Eunji Bea,
  • Dong Eun Lee,
  • Dong Hyeon Kang,
  • Rang Jin Yoo

摘要

The increasing demand for high functionality and dimensional precision in manufacturing has intensified the need for advanced production technologies for corrosion-resistant stainless-steel fasteners. Conventional nut production relies largely on cold forging or machining; however, stainless steels such as SUS304 and SUS316 exhibit low formability at room temperature, resulting in machining-dependent processes with high cost and low material yield. To address these limitations, this study develops an intelligent warm-forging process system capable of producing stainless-steel nuts with improved precision and process stability, based on real-time multi-sensor data and AI modeling. Key process variables—including Material Heating Temperature, Wedge Stroke, Material Feed Length, Die Temperature, Product Temperature, Forming Force, and dimensional responses such as Head Outer Diameter, Body Outer Diameter, Product Height—were continuously collected from an operating production line. Statistical analysis, correlation analysis, causal inference, and machine-learning models were applied to quantify the effects of these variables on dimensional variation and defect rates. The resulting models were integrated into a smart-manufacturing execution system (MES) that supports real-time monitoring, AI-based defect prediction, and parameter guidance. A formal causal analysis using directed acyclic graphs (DAGs) and the DoWhy framework was conducted to distinguish causal from correlational relationships among process variables. Model generalization was assessed through temporally ordered train–test splitting and 5-fold cross-validation, and a systematic error-source analysis was performed to identify unmeasured confounders limiting dimensional prediction performance. The proposed framework demonstrates the feasibility of AI-assisted warm forging, enabling more consistent product quality, reduced deviation, and enhanced process reliability, thereby contributing to the digital transformation of Korea’s core manufacturing sectors.